Gene Expression of the Mouse Corneal Crystallin Aldh3a1: Activation by Pax6, Oct1, and p300

Abstract

purpose. Aldehyde dehydrogenase 3a1 (Aldh3a1) represents ∼50% of the water-soluble protein of the mouse corneal epithelial cells and thus, by analogy with the abundant lens crystallins, is considered a corneal crystallin. This study was conducted to examine the developmental pattern and transcriptional activation of Aldh3a1 gene expression in the mouse cornea.

methods. Aldh3a1 mRNA and protein were analyzed by quantitative (Q)-PCR and Western immunoblot analysis. Functional promoter analysis was examined by cotransfecting plasmids containing variable portions of the Aldh3a1 promoter fused to the luciferase reporter gene into COS-7 cells with selected transcription factors. Transcription factor binding sites were identified by electrophoretic mobility shift assays (EMSA) and chromatin immunoprecipitation assays (ChIP). In situ hybridization and immunohistochemistry were used to assess expression of Aldh3a1, Pax6, and Oct1 in the cornea.

results. Aldh3a1 expression is temporally regulated in the cornea beginning at birth and increasing 100-fold by 6 weeks of age. Pax6, Oct1, and p300 synergistically activate the Aldh3a1 promoter ∼116-fold. One Pax6 and two Oct1 binding sites were identified in vitro and in vivo in the Aldh3a1 promoter fragment analyzed. Pax6 and Oct1 are both present in the nuclei of corneal epithelial cells of the 6-week-old mouse. Finally, a reduction of Aldh3a1 correlated with reduced Pax6 in the corneas of heterozygous Small eye Pax6^+/− mice.

conclusions. Pax6, Oct1, and p300 activate gene expression of the corneal crystallin Aldh3a1 in the mouse. These transcription factors are also implicated in the high expression of crystallin genes in the lens, consistent with the “refracton hypothesis” unifying many aspects of the lens and cornea.

The transparent lens and cornea share many developmental, biophysical, and physiological attributes. During embryogenesis, both tissues mature from the same thickened, ectodermal layer known as the lens placode, ^{1 2 3 4 5} through an overlapping set of transcription factors including Pax6, Sox 2, Foxe3, Pitx2, and Oct1. ^{6 7 8 9 10 11 12 13} Located at the anterior surface of the eye, the two structures refract light onto the retina enabling focused image formation, an essential step in vision. Further, the lens and cornea constitute a protective barrier, preventing adverse environmental insults such as solar radiation and pollutants (including oxygen) from reaching the internal structures of the eye. ^{14 15} The transparent, refractive, and protective properties of the lens and cornea are subserved by water-soluble, abundant proteins known as crystallins. ^{16 17 18 19 20} The crystallins (often enzymes and stress proteins) have numerous roles in the lens and cornea, a situation termed gene-sharing, to characterize the use of the same gene for more than one molecular function. ^{18 21} The similarities between lenses and corneas have led to the “refracton hypothesis,” which proposes that the lens and cornea be viewed as a single functional unit. ²²  

Corneas accumulate high proportions of a few water-soluble proteins which, by analogy with lens crystallins, are called corneal crystallins. For example, 5% to 50% of the water-soluble protein of the corneal epithelial cells of most mammals is aldehyde dehydrogenase 3a1 (Aldh3a1). ^{23 24 25 26} Aldh3a1 is present at much lower amounts in other tissues. Also like lenses, the identity of the corneal crystallin can vary from species to species, a phenomenon referred to as taxon-specificity. ^{18 27 28} Therefore, in zebrafish and the four-eyed fish Anableps, a gelsolin-like protein ^{29 30} recently identified as a scinderin paralog ³¹ represents 50% of the water-soluble protein of the adult corneal epithelial cells. By contrast, isocitrate dehydrogenase, ³² peptidyl-prolyl cis-trans isomerase, ²⁷ and glutathione S-transferase ²⁷ are corneal crystallins in cows, chickens, and squid, respectively. Finally, corneal crystallins perform multiple functions, as do lens crystallins. ^{15 18 21 33} Aldh3a1 protects against oxidative damage via multiple pathways. ¹⁵ As a member of the ALDH enzyme family, Aldh3a1 metabolizes toxic aldehydes produced by light-induced lipid peroxidation with a concomitant generation of antioxidant cofactor, NADPH. Aldh3a1 has been implicated in direct absorption of UV light, ^{23 34} scavenging reactive oxygen species, ^{35 36} acting as a chaperone, ^{37 38} and lengthening the cell cycle. ³⁹ Although the corneas of Aldh3a1 null mice are transparent, ²⁶ the eyes are more susceptible to UV-induced cataract and corneal opacity than those of wild-type mice. ⁴⁰  

The developmentally regulated, tissue-preferred, high expression of lens crystallins is orchestrated through the interplay of tissue-restricted and ubiquitously expressed transcription factors in concert with coactivator/chromatin remodeling proteins acting on distinct, but often similar combinations of DNA elements. ^{41 42} The importance of tissue-restricted transcription factors from the Pax family is underscored by their regulation of crystallin promoters from invertebrates to vertebrates. ^{41 43 44 45 46 47 48 49 50 51 52} Other transcription factors with restricted patterns of expression that independently or together with Pax6 act to regulate lens crystallin expression include members of the Maf, ^{49 51 52 53 54 55 56} SOX, ^{49 55 57} and retinoic acid receptor ^{47 56 58} families, as well as Prox1. ^{55 59} Ubiquitously expressed factors such as CREB, ^{48 52 60 61 62} TFIID, ⁶³ pRb, ^{42 63} and USF ^{60 64 65} combine with the tissue-restricted factors and with the coactivators p300 ⁵⁵ and ASC-2 ⁶⁶ to produce an abundance of crystallins in the lens. 

Much less is known about the tissue-specific regulation or the developmental expression pattern of corneal crystallins. In the current study, we found that mouse Aldh3a1 gene expression was temporally regulated in the cornea with substantial increases in mRNA and protein detected by postnatal day (PN) 14. We uncovered evidence that Pax6 and p300, regulatory proteins for lens crystallin genes, also control expression of the corneal crystallin gene, Aldh3a1. Finally, we noted that Oct1 (also known as Pou2f1), a ubiquitously expressed, POU domain-containing protein, acted synergistically with Pax6 and p300 to activate the Aldh3a1 promoter. 

Methods

mRNA Determinations

For Q-PCR, RNA was isolated from C57/BL6 mouse total cornea at PN0 (newborn), PN6, PN9, PN14, 3 weeks, and 6 weeks. Two micrograms of corneal RNA (RNA-B; Tel-Test, Friendswood, TX) was used to synthesize cDNA (High Capacity cDNA Archive Kit; Applied Biosystems, Inc. [ABI], Foster City, CA) with random primers. Q-PCR was performed by using PCR master mix (TaqMan Universal master mix; ABI) and ALDH3a1 (Mm00839312_m1) primers from gene-expression assays (TaqMan assays; ABI). Levels of eucaryotic 18S rRNA (4352930E) remained steady over the developmental time examined in this study and, therefore, was used as an endogenous control. Assays were performed with a sequence-detection system (model 7900HT; ABI). Total corneal RNA was extracted from individual, heterozygous Small eye (Sey) mice containing a deletion in the Pax6 gene (Pax6 ^SeyDey ). ^{67 68} cDNA was prepared from the entire RNA sample by using oligo dT (SuperScript First-Strand Synthesis System; Invitrogen, Carlsbad, CA). Aldh3a1 mRNA levels were determined by Q-PCR as just described, and sample size was normalized using GAPDH3 primers (FAM-MGB4352662-0506003). Mice were handled in accordance with the guidelines set forth by the Animal Care and Use Committee of the National Eye Institute, National Institutes of Health, and the ARVO Statement for the Use of Animals in Ophthalmic and Vision Research. 

Western Blot Analysis

Corneas were dissected from eyes of C57Bl/6 mice at PN0, PN6, PN9, PN14, PN21, and 6 weeks of age and the epithelial cell layer was separated from the stroma and endothelium by treatment with Dispase (10 mg/mL; Roche, Indianapolis, IN) for 30 minutes at 37°C. Whole-cell corneal epithelial lysates were prepared as described previously. ⁶⁹ Equivalent amounts of protein (2.5 μg) were separated on 12% Bis-Tris precast gels (Invitrogen), transferred to a PVDF membrane, incubated with α-Aldh3a1 antibody (1:7500; Ronald G. Lindahl, Sanford School of Medicine, University of South Dakota, Vermillion, SD), and visualized with a secondary α-rabbit HRP antibody (1:10,000; Jackson Immunochemicals, West Grove, PA) reacted with ECL (GE Healthcare, Piscataway, NJ), according to the manufacturer’s instructions. 

Aldh3a1 Promoter Constructs

Mouse Aldh3a1 promoter fragments were constructed by PCR with a mouse Aldh3a1 clone containing sequences from −1050 to +3486 that was obtained previously. ⁷⁰ For a list of oligonucleotides used to generate the Aldh3a1 promoter constructs see Supplementary Table S1. The amplified PCR promoter fragments were digested with the appropriate restriction enzymes and subcloned into the pGL3 basic vector upstream of the firefly luciferase gene (Promega, Milwaukee, WI). Large-scale plasmid DNAs were then prepared (Qiagen, Valencia, CA). The final constructs were verified by sequencing and/or PCR analysis. 

The Pax6 homeodomain-binding site predicted to reside at approximately position −980 in the Aldh3a1 promoter was mutated from 5′-taatcttatta-3′ to 5′-tCCtcttaCCa-3′, with a mutagenesis kit used according to the manufacturer’s instructions (Quick-change; Stratagene, La Jolla, CA). 

Transfections and Reporter Assays

COS-7 cells were transfected (Fugene; Roche) with 0.5 μg Aldh3a1 promoter construct plus 2 ng pSV40RL/well (to normalize for transfection efficiency) of a 12-well plate in triplicate. Various amounts of transcription factor DNAs or pKW vector only were cotransfected per well. After 48 hours, the cells were harvested, and luciferase activity was determined in triplicate, using the dual luciferase assay (Promega) according to the manufacturer’s instructions. The multiples of change (x-fold) in luciferase activity were calculated by normalizing for transfection efficiency, cotransfection with pKW vector only, and luciferase activity obtained after transfection of the appropriate promoter vector without added transcription factor. Experiments were performed on at least three separate occasions. 

An expression vector encoding Pax6 in CMV-driven vector pKW10 has been described previously. ^{71 72} Expression constructs were generated for mouse Kruppel-like factor (Klf) 4, ets-like factor 3 (Elf3), interferon regulatory factor (Irf) 1, Mafk, and Mafg by PCR with adult mouse corneal cDNA and the oligonucleotides listed in Supplementary Table S2. The cDNAs were subcloned into the pCI vector containing the CMV promoter (Promega) and sequenced. 

Clones for cold-shock domain protein A (Csda) (MMM1013-9200235), doublesex and mAb-3-related transcription factor 5 (Dmrta2) (EMM1002-1727147), grainyhead like 2 (Tcfcp213) (MMM1013-9201448), Irf7 (EMM1002-441), B-cell leukemia/lymphoma 3 (Bcl3) (EMM1002-21668), and myc-associated zinc finger protein (Maz) (EMM1002-411) were purchased from OpenBiosystems (Huntsville, AL). Clones for Klf5 (MGC-13705), Junb (MGC-6021), zinc finger protein 191 (Zfp191; MGC-19271), PDZ and LIM domain 1 (Pdlim1; MGC-5634), max dimerization protein (Mad; 3390903), pou2f1 (Oct1; 10000140), ets variant gene 3 (Etv3; 9145152), tumor susceptibility gene 101 (Tsg101; 9154444), sterol regulatory element binding factor 2 (Srebf2; 9848039), odd-skipped related protein (Osr2; MGC-19,171), Irf6 (MGC-5918), Irf1 (MGC-6190), transcription factor 12 (Tcf12; 9990943), fos-related protein 2 (Fosl2; 10007154), MafK (MGC-14,027), and ets homologous factor (Ehf; MGC-8140) were purchased from ATCC (Manassas, VA). A clone for jun dimerization protein 2 (Jundm2; 4012135) and p300 (21-178) was purchased from Invitrogen and Upstate Biotechnology (Lake Placid, NY), respectively. 

Electrophoretic Mobility Shift Assay

Sequences of the oligonucleotides corresponding to potential Pax6 sites within the Aldh3a1 promoter are shown below. The region of the mouse Aldh3a1 promoter corresponding to each oligonucleotide is indicated in parentheses. Oligo #1 (−115 to −91) top strand (TS): 5′-aggtctggattatgctgcacccaca-3′ and bottom strand (BS): 5′-tgtgggtgcagcataatccagacct-3′; oligo #2 (−100 to −76) TS: 5′-tgcacccacatttgaatattgtttt-3′ and BS: 5′-aaaacaatattcaaatgtgggtgca-3′; oligo #3 (−85 to −61) TS: 5′-atattgttttatcttcgtcacgcga-3′ and BS: 5′-tcgcgtgacgaagataaaacaatat-3′; oligo #4 (−70 to −46) TS: 5′-cgtcacgcgaacttcctggggaagg-3′ and BS: 5′-ccttccccaggaagttcgcgtgacg-3′; oligo #5 (−65 to −41) TS: 5′-cgcgaacttcctggggaaggactct-3′ and BS: 5′-agagtccttccccaggaagttcgcg-3′; oligo #6 (consensus Pax6) TS: 5′-tcgaggtgcactcatgcgtgaagtcga-3′ and BS: 5′-tcgacttcacgcatgagtgcacctcga-3′; and oligo #7 (mutated consensus Pax6) TS: 5′-tcgaggtggtctcatcagtgaagtcga-3′ and BS: 5′-tcgacttcactgatgagaccacctcga-3′. Sequences of the oligonucleotides corresponding to potential Oct1 sites within the Aldh3a1 promoter are as follows. The region of the Aldh3a1 promoter corresponding to each oligonucleotide is indicated in parentheses. oligo #8 (Oct1 consensus site) TS: 5′-tgtcgaatgcaaatcactagaa and BS: 5′-ttctagtgatttgcattcgaca-3′; oligo #9 (+3010 to +3034) TS: 5′-gaattttgatgctagtttttcttta-3′ and BS: 5′-taaagaaaaactagcatcaaaattc-3′; oligo #10 (+3054 to +3078) TS: 5′-catggtctgtatgcatccaacaact-3′ and BS: 5′-agttgttggatgcatacagaccatg-3′: oligo #11 (+3311 to +3335) TS: 5′-cggtgggaatgcattttccagtctc-3′ and BS: 5′-gagactggaaaatgcattcccaccg-3′; oligo #12 (+3434 to +3458) TS: 5′-tccttcctctgatgcaaagtgttct-3′ and BS: 5′agaacactttgcatcagaggaagga-3′; oligo #13 (−1109 to −1086) TS: 5′-agagcaaatgcaaaaggcactaagtc-3′ and BS: 5′-gacttagtgccttttgcatttgctct-3′; and oligo #14 (−803 to −779) TS: 5′-gaactgaaatgcaaaaacagtgggt-3′ and BS: 5′-acccactgtttttgcatttcagttc-3′. oligonucleotides (100 ng each) were end-labeled with γ[³²P]ATP, annealed to form a double-stranded (ds) duplex, and purified on a G50 column (specific activity, >500,000 cpm/μL). Nuclear extracts (8–10 μg) from αTN4 cells for Pax6 analysis, prepared, as described previously, ⁷³ or HeLa cells (Promega) for Oct1 analysis were combined with 5× EMSA (electrophoretic mobility shift assay) buffer (Promega) for 10 minutes at room temperature, followed by the addition of 200,000 cpm labeled probe for 20 minutes at room temperature. The resultant complexes were analyzed on a 6% nondenaturing retardation gel (Invitrogen). Competition assays were performed by adding unlabeled, oligonucleotide duplex (50 and 100 ng) for 10 minutes at room temperature before adding the labeled probe. For EMSAs, 1 μL α-Pax6 antibody (PRB-278P; Covance, Vienna, VA) or α-Oct1 antibody (sc-232; Santa Cruz Biotechnology, Santa Cruz, CA) was added (for 1 hour) after the addition of labeled ds oligonucleotides. Nonspecific antibody controls were performed with α-Neu Ab (sc-284; Santa Cruz Biotechnology). 

ChIP Analysis

NCTC cells, a line derived from human skin that expresses Aldh3a1 endogenously, was transfected with Pax6 and used as the source of chromatin. Proteins were cross-linked using formaldehyde (36.5% Solution; Sigma-Aldrich, St. Louis, MO), cells were lysed with a chromatin immunoprecipitation (ChIP) assay kit (Upstate Biotechnology) and the chromatin was sheared by sonication according to the manufacturer’s instructions. Three aliquots (100 μL each: ∼ 2 × 10⁶ cell equivalents) of sheared, cross-linked chromatin were incubated with 1 μg of rabbit α-Pax6 antibody (PRB-278P; Covance), goat α-Pax6 antibody (sc-7750; Santa Cruz Biotechnology), or rabbit α-Neu antibody (sc-284; Santa Cruz Biotechnology) at 4°C overnight. One percent of the supernatant fraction was reserved as input chromatin for PCR analysis. Immunoprecipitation reactions were washed, the chromatin was eluted from the beads, and DNA-protein cross-links in all samples including the input were reversed according to the manufacturer’s instructions (Upstate Biotechnology). DNA was isolated by phenol-chloroform extraction and ethanol precipitation, and 5 μL of each 50-μL sample was used in a PCR reaction. The PCR primer sequences, corresponding to the −146/+4 region of the human Aldh3a1 promoter, were ALDH3a1 −146 TS: 5′-cctgggctgtagggagcagaggtc-3′ and ALDH3a1 +4 BS: 5′-ccaagaggggacgtatttaaggac-3′. The region between −2537 and −2337 of the Aldh3a1 promoter was amplified with the following primers: ALDH3a1 −2537 TS: 5′-ctaactcatggctgatgctaac-3′ and ALDH3a1-2337 BS: 5′-ctcttcccacttgacatttattac-3′ for an unrelated ChIP control. PCR conditions were as follows: 5 minutes at 94°C followed by 34 cycles of 20 seconds at 94°C, 30 seconds at 55°C, 30 seconds at 72°C, and a final extension for 2 minutes at 72°C. PCR products were separated by 1% TBE gel electrophoresis and visualized by ethidium bromide staining. 

Six week-old mouse corneal epithelial extracts were used as a source of chromatin for ChIP analysis of Oct-1 binding sites. Aliquots of cross-linked chromatin were incubated with 1 μg of rabbit α-Oct1 (sc-232; Santa Cruz Biotechnology) or rabbit α-Neu (sc-284; Santa Cruz Biotechnology) antibody and processed as described earlier. The PCR primer sequences corresponding to the mouse Aldh3a1 promoter at positions +3342/+3486 were ALDH3a1 +3342 top strand (TS): 5′-agtctttatccaaataaaattagc-3′ and ALDH3a1 +3486 bottom strand (BS): 5′-catggtaactggggatagagaac-3′, at −850/−650 were ALDH3a1 −850 TS: 5′-ggagagcaagaacacaggcttgg-3′ and ALDH3a1 −650 BS: 5′-gtccctttctgtatgtttagcc-3′, and unrelated control primers described earlier. 

In Situ Hybridization, Immunohistochemistry, and Immunofluorescence

Freshly frozen, 10-μm sections of C57/Bl6 6-week-old mouse eyes were processed for immunohistochemistry, as described previously. ⁷⁴ Paraformaldehyde-fixed sections were incubated overnight at 4°C degrees with α-Pax6 Ab (1:600; PRB-278P; Covance) or α-Oct1 Ab (0.2 ng/μL sc-232; Santa Cruz Biotechnology). The α-Oct1 Ab was preincubated for 2 hours at room temperature with a nonspecific EGFR peptide (1 ng/μL sc-03P; Santa Cruz Biotechnology) or a specific Oct1 peptide (1 ng/μL sc-232P; Santa Cruz Biotechnology). For negative controls, primary antibody was omitted during primary antibody incubation. Immunolocalization of Pax6 and Oct1 were visualized (Elite Vectastain kit and DAB kit; Vector Laboratories, Burlingame, CA), according to the manufacturer’s instructions. Localization of Oct1 was also examined by immunofluorescence. Corneal sections were treated with Oct1 Ab prepared as described earlier, reacted with Alexa Fluor 568 secondary antibody (1:250 in 10% normal goat serum; Invitrogen-Molecular Probes), and counterstained with DAPI (4′,6′-diamino-2-phenylindole), as described previously. ⁷⁴ Finally, the slides were mounted in anti-fade reagent (Invitrogen-Molecular Probes) and viewed under a fluorescence microscope (AxioPlan 2; Carl Zeiss Meditec, Inc., Thornwood, NY) equipped with a CCD camera for epifluorescence (AxioVision; Carl Zeiss Meditec, Thornwood, NY). 

Riboprobes were synthesized (DIG RNA Labeling Kit [Sp6/T7]; Roche) with linearized, proteinase-K-treated, full-length plasmid cDNAs for Aldh3a1. Sense or antisense riboprobe/mL hybridization buffer (200 ng) was hybridized to adult, cryopreserved corneal sections from Pax6 ^SeyDey mice and Pax6 ^SeyNeu mice that contain a point mutation in the Pax6 gene, at 55°C. Processing and detection of digoxigenin-labeled hybridization was as described previously. ⁶⁹ A morphologic assessment of the enucleated eyes was used to establish the genotype of each mouse. 

Results

Developmental Regulation of Aldh3a1 Expression in the Mouse Cornea

Aldh3a1 mRNA levels have been estimated to comprise up to 1% of the total mRNA in the adult mouse cornea. ⁷⁵ To elucidate the temporal pattern of Aldh3a1 gene expression in mouse cornea, mRNA levels were assessed by Q-PCR. No Aldh3a1 mRNA was detected at embryonic day (E)17 in the whole eye (data not shown). Aldh3a1 mRNA was first detected in total cornea at PN0 and increased 100-fold by 6 weeks of age (Fig. 1A) . A steep increase in Aldh3a1 mRNA of ∼15-fold occurred between PN9 and PN14, coincident with eye opening (Fig. 1A) . 

The developmental pattern of Aldh3a1 protein expression was evaluated by Western blot analysis of corneal epithelial extracts from different postnatal times. Aldh3a1 protein was first detected at PN9 with an anti-Aldh3a1 antibody (Fig. 1B) . Consistent with the Aldh3a1 mRNA levels, there was a substantial increase in the amount of Aldh3a1 protein between PN9 and PN14 (Fig. 1B) . In general, the developmental expression of Aldh3a1 protein paralleled that of the mRNA. 

Transcription Factor Regulation of the Aldh3a1 Gene

To identify candidate transcription factors present in the developing cornea that function in Aldh3a1 gene regulation, we surveyed the expression data compiled from a SAGE analysis of PN9 and 6-week-old mouse corneas. ⁷⁵ A subset of transcription factors that were either cornea-enriched (at least three times more in the cornea than in eight other normal mouse SAGE libraries) or developmentally regulated (at least a twofold change from PN9 to 6 weeks) were considered to be candidates and are shown in Table 1 . 

Cotransfection experiments were performed in COS-7 cells with a mouse Aldh3a1 promoter fragment–reporter gene construct referred to as −1050/+3486 Luc. COS-7 cells, a kidney cell line, were chosen to minimize interference by endogenous transcription factors that are important for the expression of corneal crystallin genes that might be present in a corneal cell line. A similar Aldh3a1 promoter fragment fused to the chloramphenicol acetyltransferase reporter gene shows cornea-preferred activity in transgenic mice. ⁷⁰ The −1050/+3486 Luc construct contains 1050 base pairs upstream of the transcription start site, the noncoding exon 1, an ∼3500-bp intron, and part of exon 2 fused to the firefly luciferase (Luc) reporter gene (Fig. 2) . Basal Aldh3a1 promoter activity was weak (∼2 times above a promoterless vector) in COS-7 cells; however, cotransfection with several transcription factors resulted in an increase in promoter activity. Maximum amounts (60 ng) of Pax6, Oct1, Klf4, or Klf5 stimulated the Aldh3a1 promoter 9.1-, 46-, 9.1-, and 24-fold, respectively (Table 1) . Modest Aldh3a1 promoter activation of between three- and sixfold was observed with Elf3, Mafg, Jun, Irf7, or a combination of Maz and Mxd1. The other transcription factors tested showed a negligible effect on the promoter activity. Of interest, although IRF1 alone did not alter promoter activity, it reduced activation of the promoter by Pax6 (data not shown). 

Synergistic Activation of the Aldh3a1 Promoter by Pax6, Oct1, and p300 In Vitro

We focused our investigation of the regulation of the Aldh3a1 promoter on Pax6, previously identified as an important regulator of crystallin gene expression in the lens, and on Oct1, the most effective activator of the Aldh3a1 promoter fragment that we tested. First, the ability of increasing amounts of Pax6 or Oct1 to activate the −1050/+3486 Luc Aldh3a1 promoter was examined by cotransfection in COS-7 cells. Pax6 (Fig. 3A)and Oct1 (Fig. 3B)activated the promoter 11.1- and 41-fold, respectively, at the highest amount of transcription factor used (60 ng). To test for a possible functional interaction between Pax6 and Oct1, we conducted transient cotransfection experiments. Separate Western immunoblot experiments showed that the nontransfected COS cells lacked Pax6 and Oct1 and that transfection by Pax6 did not induce Oct1 or vice versa (data not shown). Coexpression of Pax6 with Oct1 enhanced promoter activation at all doses examined (Fig. 3C) , reaching a 116-fold activation with maximum amounts of combined transcription factors (third set of bars from the left in Fig. 3C ). We consider the coactivation of the Aldh3a1 promoter by Pax6 and Oct1 kinetically synergistic, as defined Herschlag and Johnson. ⁷⁶  

p300, a well known transcriptional coactivator, interacts with many transcription factors including those important in lens crystallin gene expression. ^{55 77} To determine whether p300 may be involved in coactivation of the Aldh3a1 promoter by Pax6 and Oct1, an expression plasmid encoding p300 was introduced with p−1050/+3486 Luc into the COS-7 cells by itself, or with expression plasmids expressing Oct1, Pax6, or both Pax6 and Oct1. Consistent with reports that p300 is a coactivator and does not directly bind DNA, p300 alone resulted in only a modest increase in Aldh3a1 promoter activity (Fig. 3C) . Cotransfection of p300 with Oct1 did not alter activity over what was observed for Oct1 alone. In contrast, cotransfection of p300 with Pax6 enhanced activation of the Aldh3a1 promoter beyond the sum of the effects of each transcription regulator alone (Fig. 3C) . Finally, combining p300 with Pax6 and Oct1 (60 ng each) resulted in a 254-fold induction in Aldh3a1 promoter activity, an enhancement that surpasses the sum of each of the two-factor combinations (Fig 3C , rightmost histogram). In summary, these results indicate synergistic activation of theAldh3a1 promoter by Pax6 and Oct1, by Pax6 and p300, and especially by a combination of the three transcriptional regulators. 

Binding of the Aldh3a1 Promoter by Pax6 and Oct1 In Vitro and In Vivo

To delineate region(s) of the Aldh3a1 promoter where Pax6 exerts its action, we cotransfected plasmids containing deletions from the 5′ and 3′ ends of the −1050/+3486 Aldh3a1 promoter fragment with Pax6 (see Fig. 2for a schematic of constructs) into COS-7 cells. Removal of the first intron, from positions +2 to +3486, resulted in an approximate twofold reduction in luciferase activity with Pax6 (30 ng) compared with the −1050/+3486 Luc plasmid (Fig. 4A) . Inspection of the intronic sequence, however, did not reveal a consensus Pax6-binding site. A similar search in the −1050/+1 region of the promoter fragment revealed a Pax6 homeodomain consensus binding site at positions −975 to −965. ⁷⁰ Mutation of 4 of the 11 bases of this putative homeodomain binding site did not alter Pax6-inducible luciferase activity, indicating that this site is not used by Pax6 under these conditions (data not shown). Corroborating this result, Pax6 activation of the Aldh3a1 −400/+1 promoter fragment was the same as for the −1015/+1 promoter fragment indicating that the sequences between −1050 and −400 are not necessary for Pax6 induction. In contrast, a modest but reproducible twofold reduction in promoter activity occurred with the removal of bases from positions −110 to −50 (Fig. 4A) . Inspection of this short region revealed a potential Pax6 paired-domain binding site from −78 to −65. An optimal alignment between this sequence and the consensus Pax6 site showed a single base mismatch (Fig. 4B) . ⁴⁸  

To assess whether Pax6 binds directly to the DNA, we incubated a series of overlapping, ds ³²P-labeled oligonucleotides corresponding to the region between −115 and −41 of the Aldh3a1 promoter with αTN4 cell nuclear extracts, a lens cell line that expresses Pax6 endogenously and analyzed by EMSA. Oligonucleotides #3 (Fig. 5A , lane 3) and #4 (Fig. 5A , lane 4), both containing the 3′ half of the conserved Pax6 binding site, were shifted to the same position on the gel as the Pax6 consensus oligonucleotide #6 (Fig. 5A , lane 6). The specificity of each complex was analyzed by addition of excess, unlabeled ds oligonucleotides (50 and 100 ng). Nonradioactive oligonucleotides #3 (Fig. 5B , lanes 4, 5) or #4 (Fig. 5B , lanes 6, 7) were able to compete with the #6 radiolabeled consensus Pax6 oligonucleotide and nonradioactive oligonucleotide #6 was likewise able to compete with radiolabeled oligonucleotides #3 (Fig. 5B , lanes 15, 16) and #4 (Fig. 5B , lanes 22, 23). Incubation of the binding reaction with a Pax6 antibody, but not an unrelated Ab (Fig. 5C , lanes 2, 5, 8), eliminated the formation of the complex on the control # 6 oligonucleotides (Fig. 5C , lane 3) and reduced complex formation using oligonucleotides #3 (Fig. 5C , lane 6) or #4 (Fig. 5C , lane 9), consistent with Pax6 binding. 

To investigate whether Pax6 binds directly to the Aldh3a1 promoter in vivo, we performed a ChIP assay using two different, but specific, antibodies against Pax6 to immunoprecipitate chromatin from human NCTC skin cells. NCTC cells were chosen because it is the only cell line, to our knowledge, that expresses Aldh3a1 endogenously. ³⁹ Because of the extreme difficulty of isolating sufficient quantities of corneal tissue for ChIP, we considered the NCTC cells to be the best alternative. The human Aldh3a1 sequence differs from the mouse Aldh3a1 sequence in the region between −78 and −65 by 1 bp. DNA products of the right size (142 bp) were amplified from reverse cross-linked, goat or rabbit α-Pax6-immunoselected DNA with primers complementary to −146 and −4 of the human promoter, demonstrating that Pax6 binds to the human Aldh3a1 promoter within this region (Fig. 5D , lanes 2, 3, respectively) whereas no product was detected with a nonspecific promoter region −2537 to −2337 (data not shown). When an unrelated α-Neu antibody was used for ChIP analysis, no amplification product was observed (Fig. 5D , lane 4). A lower migrating band appears in all lanes and may represent nonspecific amplification of primer dimers. We conclude that Pax6 interacts directly with the mouse Aldh3a1 promoter between positions −85 and −46, in vitro, and with the human Aldh3a1 promoter between positions −146 and −4, in vivo. 

A similar strategy was used to identify the Oct1-binding region(s) on the Aldh3a1 promoter. Plasmid DNAs containing deletions from the 5′ and 3′ ends of the −1050/+3486 Aldh3a1 promoter fragment fused to the luciferase gene were cotransfected with the Oct1 expression plasmid (30 ng). Promoter activation dropped ∼65% compared with the activity of the −1050/+3486 promoter plasmid when sequences at the 3′ end of the first intron, from +2986 to +3486, were removed (Fig. 6A) . Four potential Oct1 binding sites were found within this region and are aligned with the consensus Oct1-binding site in Figure 6B . ⁷⁸ The −1050/+1 Aldh3a1 promoter construct was as active as the one containing the −1050/+2986 promoter fragment suggesting that no other Oct1 sites exist in the first intron. A threefold reduction in luciferase activity occurred when bases from −1050 to −500 were deleted (Fig. 6A) . One putative Oct1 binding site was identified between −1050 and −50 (Fig. 6B) . 

To assess whether Oct1 binds directly to the DNA, ds ³²P-labeled oligonucleotides corresponding to the five putative Oct1 sites were incubated with HeLa cell nuclear extracts. The resultant complex(es) were analyzed by EMSA. Each of the oligonucleotides resulted in a complex that migrated to the same position as the Oct1 consensus oligonucleotide complex (Fig. 7A) . The specificity of each complex was analyzed by addition of excess, unlabeled ds oligonucleotides. Sites corresponding to oligonucleotides #9 and #10 were eliminated as possible Oct1 binding sites as nonradioactive oligonucleotides failed to compete with their respective radiolabeled oligonucleotides (data not shown). Excess, unlabeled oligo #12 (Fig. 7B , lanes 6, 7) or #14 (Fig. 7B , lanes 9, 10), but not #11 (Fig. 7B , lanes 4, 5) competed with radiolabeled Oct1 consensus oligonucleotides for binding. In the complementary experiment, excess unlabeled Oct1 consensus oligonucleotides competed with radiolabeled oligos #12 (Fig 7B , lane 15) and #14 (Fig 7B , lane 19). Incubation of the binding reaction with an α-Oct1 antibody (Fig. 7C , lanes 2, 5, 8, 11), but not an unrelated Ab (Fig. 7C , lanes 3, 6, 9, 12), eliminated the complex on the consensus octamer site, oligos #12, and #14, but not on an unrelated oligonucleotide complex, suggesting that Oct1 binds directly to sites within oligos #12 and #14. Oligos #12 and #14 correspond to positions +3434 to +3458 and −803 to −779 in the mouse Aldh3a1 promoter, respectively. 

To investigate whether Oct1 binds directly to the Aldh3a1 promoter in vivo, we performed a ChIP assay using a specific α-Oct1 antibody to select chromatin from adult mouse corneal epithelial extracts. Products of the right size were amplified from reverse cross-linked Oct1-immunoselected DNA, with primers complementary to positions −850 and −650 (Fig. 7D , lane 3) and +3342 and +3486 (Fig. 7E , lane 3) of the mouse promoter, demonstrating that Oct1 binds to the mouse Aldh3a1 promoter within these regions. Much less amplification product was observed with an unrelated antibody (Figs 7D , lane 2; 7E, lane 2), and no product was detected with a nonspecific promoter region, −2537 to −2337 (data not shown). From these results, we conclude that Oct1 interacts directly with the mouse Aldh3a1 promoter in the 5′ flanking region and in the first intron, in vitro and in vivo. 

Localization of Pax6 and Oct1 to Nuclei of Corneal Epithelial Cells

To verify the relevance of our findings in COS-7 cells to Aldh3a1 expression in vivo, we tested whether Oct1 protein is expressed in the same location as Pax6 in the mouse cornea. Immunohistochemistry performed on 6-week-old mouse corneal cryosections showed strong nuclear expression of Pax6 in the epithelial cells (Fig. 8B) , as previously reported. ⁷⁹ Strong Oct1 immunostaining was observed throughout corneal epithelial cells treated with α-Oct1 antibody preincubated with a nonspecific peptide (Fig. 8D) . The Oct1 immunostaining was lost when the α-Oct1 Ab was preincubated with the peptide by which it was generated (Fig. 8E) , indicating that the staining observed in Figure 8Dis specific for Oct1. In addition, weaker staining for Oct1 was noted in stromal keratocytes (Fig. 8D) . 

The expression of Oct1 throughout the epithelial cells as determined by immunohistochemistry suggests that Oct1 is present in the cytoplasm as well as the nucleus, as has been shown. ⁸⁰ To determine definitively whether Oct1 is located in the nucleus, we used double fluorescence coupled with confocal imaging. The nuclei were visualized with DAPI staining (Fig. 8G) , whereas Oct1 was localized by red immunofluorescence (Fig. 8F) . When the two images were superimposed, most of the nuclei appeared purple, indicating the presence of Oct1 in the nucleus (Fig. 8H) . 

Correlation of Reduction of Pax6 mRNA with a Reduction of Aldh3a1 mRNA In Vivo

We tested the ability of Pax6 to regulate Aldh3a1 mRNA expression in vivo using Pax6 ^SeyDey and Pax 6 ^SeyNeu mice, two strains of Small eye (Sey) mice with a deleted or mutated Pax6 gene, respectively (see the Methods section). First, mRNA levels for Aldh3a1 were examined in the two strains of mice by in situ hybridization. A robust signal for Aldh3a1 mRNA was observed in the corneal epithelium of the wild-type Pax6 ^SeyDey (+/+) mouse using an Aldh3a1 antisense riboprobe (Fig. 9A) . The Aldh3a1 mRNA hybridization signal was greatly reduced in the Pax6 ^SeyDey (Fig. 9B)and Pax6 ^SeyNeu (Fig. 9C)heterozygous (+/−) corneas that have reduced levels of Pax6 protein. The presence of K12 mRNA in the corneas of the Sey mice indicates that these cells were corneal epithelial cells and not invading conjunctival cells in the mutant mouse strains (data not shown). Q-PCR was performed to quantitate the reduction in Aldh3a1 mRNA in corneas from Pax6 ^SeyDey mice. Use of primers specific for Aldh3a1 (TaqMan; ABI) showed a 2.3-fold reduction of Aldh3a1 mRNA in corneas from Pax6 ^SeyDey (+/−) mice compared with the wild-type Pax6 ^SeyDey (+/+) mice (Fig. 9D) . GAPDH3 was used as an endogenous control for normalization purposes. Although it seems likely the corneal defects are due to reduced Pax6 dosage in the Pax6 ^SeyDey mouse, which contains a large deletion including the Pax6 gene, this has not been shown formally. Together, these results are consistent with Pax6 regulation of Aldh3a1 gene expression in the mouse cornea. 

Discussion

Despite the diversity of lens crystallins within as well as between species, a similar set of transcription factors regulates their preferentially high expression in the lens. ²¹ Pax6 has received particular attention, because it plays a critical role in lens development in vertebrates and invertebrates ^{81 82} and has been shown to be a direct regulator of crystallin gene expression. ^{41 42 43 44 45 46 47 48 49 50 51 52 55} In this study, we showed that Pax6 also regulates expression of the mouse Aldh3a1 gene in the cornea. ^{23 24 26} In addition to activation of the corneal crystallin gene, Pax6 activates noncrystallin corneal genes, ^{83 84 85 86} consistent with its importance for gene expression in cornea as well as lens. The present finding that Pax6 regulates the corneal crystallin gene for Aldh3a1, as well as virtually all lens crystallin genes tested, ^{18 21} supports the “refracton hypothesis” proposing that the lens and cornea form a functional unit with many similarities. ^{21 22} This does not mean of course that all the transcription factors used for Aldh3a1 gene expression will be the same as those used for the expression of lens crystallin genes. It is of interest that an Aldh family member, known as Ω-crystallin, has been recruited as a lens crystallin in cephalopods (squid and octopus) ⁸⁷ and scallops ²⁰ and appears to use Pax6 to activate the promoter of its gene. ⁴⁸ Thus, the Aldh3a1 gene joins a growing group of target genes for Pax6 in the eye. Eye-specific markers in addition to lens crystallins whose genes are controlled by Pax6 include rhodopsin, ⁸⁸ K12, ⁸⁴ and optimedin. ⁸⁹  

Partnering with other transcriptional regulators may modify the transcriptional activities of Pax6. Functional studies have revealed interactions between Pax6 and Sox 2, ⁴⁶ Pax6(5a), ⁷² retinoic acid nuclear receptors, ^{47 56 72} Maf family members, ^{56 72} and pRb, ⁴² resulting in synergistic activation or repression of lens crystallin promoters. Similarly in the cornea, Pax6 associates directly with AP-2α to enhance activation of the gelatinase B promoter. ⁸⁶ In the present study, we identify a new Pax6 partner, Oct1. Oct1 is a founding member of the POU transcription factor family whose members play essential functions in organ development and cellular differentiation. ⁷⁸ POU domain proteins contain a bipartite DNA-binding domain composed of a variant homeodomain and a POU-specific domain. Oct1 or Pax6 can act separately to upregulate the Aldh3a1 promoter or together, resulting in synergistic activation of the promoter. It remains to be determined whether the synergistic activity relies on a direct interaction between Pax6 and Oct1, as has been shown for Pax6 and other homeodomain-containing proteins. ⁹⁰  

The present data show that p300 can augment transactivation by Pax6, consistent with a previous report showing that transactivation of the glucagon promoter by Pax6 is enhanced by p300. ⁹¹ This well-known coactivator is also involved in lens crystallin gene expression. p300 enhanced c-Maf-induced activation of mouse αA-, βB2-, and γF-crystallin gene promoters in the lens. ⁵⁵ However, p300 did not promote Pax6 transactivation of these lens crystallin promoters, suggesting that the ability of p300 to interact with a particular transcription factor may depend on promoter context and/or availability of specific transcription factor partners. Future studies will be necessary to determine whether coactivation of the Aldh3a1 promoter by p300 is achieved through chromatin remodeling via its histone acetyltransferase activity. ⁵⁵  

Oct1 and Pax6 are both expressed in mouse corneal epithelial cells as early as PN9, making them candidates in orchestrating the upregulation of Aldh3a1 that occurs by PN14. Unexpectedly, SAGE analysis indicates that Oct1 and Pax6 mRNA levels decrease from PN9 to 6 weeks. ⁷⁵ The interpretation of decreasing Pax6 mRNA during corneal development is clouded by the facts that the SAGE tags cannot discriminate between RNAs encoding Pax6 and Pax6(5a) isoforms and that we do not know how the Pax(5a) isoform acts on the Aldah3a1 promoter. An additional consideration is that Pax6 protein can undergo posttranslational modifications that alter its transactivation properties. ⁹² For example, homeodomain-interacting kinase 1 (HIPK) phosphorylates Pax6, increasing its transactivation ability by enhancing interaction with p300. ⁷⁷ HIPKs are upregulated in the cornea from PN9 to 6 weeks of age and thus may enhance the transcriptional activities of Pax6. ⁷⁵ Similarly, the absolute amount of a transcription factor at any given stage of development does not define its involvement in the expression of a particular gene. Consequently, decreases in Pax6 and Oct1 mRNA with development in the cornea do not mean that these transcription factors are not involved in increasing Aldh3a1 gene expression. 

Although lens and corneal crystallin gene expression are developmentally regulated, the time of onset of expression differs in the two tissues. In rodents, the lens crystallins are first expressed during embryogenesis in a distinctive pattern, with α-crystallin appearing at the lens vesicle stage, followed generally by β-crystallin and then γ-crystallin. ^{41 93} In the cornea, the expression of Aldh3a1 mRNA begins at birth and increased levels of mRNA and protein are detected by PN14. The increase in Aldh3a1 mRNA and protein correlate with eye opening at approximately PN13 and the initiation of corneal epithelial stratification from 1 to 2, to 8 to 10 cell layers in thickness. It is possible that the onset of crystallin expression, whether in the lens or cornea, marks an important stage in the functional differentiation and/or maturation of their respective tissue. The accumulation of Aldh3a1 at PN13 coincides with its proposed functions involving corneal clarification, ⁹⁴ UV detoxification, ^{23 34} and regulation of cell division. ³⁹  

Certainly additional transcription factors are involved in Aldh3a1 transcriptional regulation. Indeed, regulation of eukaryotic gene expression involves the combinatorial use of multiple proteins. ⁹⁵ Other major POU factors are expressed in the cornea (Davis J, Piatigorsky J, unpublished data, 2006). Of particular interest is Skn-1a (pou2f3), thought to be selectively expressed in epidermis. By virtue of similar POU domains, Skn-1a and Oct1 have been assigned to the same class of POU factors and are expected to have similar DNA binding preferences. ⁷⁸ Of interest, Skn-1a has recently been shown to associate with Elf3, resulting in the activation of genes important in late stages of differentiation in the skin. ⁹⁶ Elf3 is also an abundant, cornea-enriched transcription factor ⁷⁵ that modestly activates the Aldh3a1 promoter by itself, but may exert greater activity in the cornea in the presence of Skn-1a and/or Oct1. A second group of transcription factors that may be involved in Aldh3a1 gene regulation are members of the SOX family. Complex formation between SOX2 and Pax6 on a lens-specific enhancer is essential for high, synergistic activation of the chicken δ-crystallin gene. ⁴⁶ Cooperative binding to the enhancer requires a Sox2 site immediately adjacent to a degenerate Pax6 site. We have found a Sox site identical with the one in the lens enhancer situated next to the Pax6 binding site identified in this study and are pursuing studies to examine a potential interaction between these two factors. It is intriguing that Sox factors have also been shown to interact with members of the POU domain family in embryonic stem cells, neural stem cells, and lens, ^{13 97 98} opening the door to the possibility that an interaction between Pax6, Oct1, and a Sox factor coordinate Aldh3a1 gene activity. The high amount of Aldh3a1 protein (∼50% of the total protein) compared with 1% of the total mRNA suggests that there are additional, posttranscriptional mechanisms regulating the expression of Aldh3a1. ^{26 75}  

Finally, Aldh3a1 appears to have several enzymatic and nonenzymatic roles in the cornea. ¹⁵ Some of these molecular functions are dependent on the high expression of Aldh3a1 in the cornea. The synergistic activation of the Aldh3a1 gene by Pax6, Oct1, and p300 in the cornea probably contributes greatly to the multiple functions of Aldh3a1. The acquisition of new protein functions by quantitative and qualitative changes in the expression of their genes, a major mechanism of gene sharing, ^{99 100} is a hallmark of lens crystallins and a general principle of evolution. ²¹ We thus conclude that the cooperative use of Pax6, Oct1, and p300 in contributing to high, cornea-preferred activity of the mouse Aldh3a1 promoter is a fundamental evolutionary cause for employment of multiple functions of Aldh3a1 via a gene-sharing strategy. 

 Submitted for publication August 13, 2007; revised November 8, 2007; accepted February 29, 2008.

 Disclosure: J . Davis, None; D . Davis, None; B . Norman, None; J . Piatigorsk y, None

 The publication costs of this article were defrayed in part by page charge payment. This article must therefore be marked “advertisement” in accordance with 18 U.S.C. §1734 solely to indicate this fact.

 

Figure 1.

 

Developmental expression of Aldh3a1 mRNA and protein in corneal epithelium. (A) Q-PCR analysis showed relative mRNA levels of Aldh3a1 in corneal epithelium isolated from mice at PN0, PN6, PN9, PN14, and 6 weeks of age (arbitrarily set at 1). (B) Western blot analysis showing Aldh3a1 protein from total cornea at the same time points.

Table 1.

 

Corneal Enriched Transcription Factors

Gene Symbol	Unigene Number	Transcription Factor	Fold Activation of Aldh3a1 Promoter	PN9 TPM	Adult TPM
Pax6	3608	Paired box gene 6	11.1	442*	96*
Klf4	4325	Kruppel-like factor 4	9.1	427	644, †
Klf5	30262	Kruppel-like factor 5	23.8	63	145
Pou2f1	245261	Oct-1	46	253	81
Jun	275071	Jun	4.4	47	0
Fosl2	24684	Fos-like antigen 2	2.1	95	64
Jundm2	103560	Jun dimerization protein 2	2.1	0	64
Irf1	105218	Interferon regulatory factor 1	<2	332	773
Irf6	305674	Interferon regulatory factor 6	<2	143	177
Irf7	3233	Interferon regulatory factor 7	3.0	32	209
Elf3	291048	E74-like factor 3	5.0	269	354
Ehf	10724	Ets homologous factor	<2	157	306
Mafk	157313	Mafk	<2	47	177
Mafg	268010	Mafg	5.3	0	32
Maz	378964	MYC-associated zinc finger protein	4.1, ‡	47	97
Mxd1	279580	Max dimerization protein	4.1, ‡	126	241
Srebf2	38016	Sterol regulatory element binding factor 2	<2	174	403
Pdlim1	5567	PDZ and LIM domain 1 (elfin)	<2	158	32
Bcl3	235309	B-cell leukemia/lymphoma 3	<2	111	0
Tsg101	241334	Tumor susceptibility gene 101	<2	47	209
Csda	299604	Cold shock domain protein A	<2	300	226
Zfp191	417427	Zinc finger protein 191	<2	190	0
Dmrta2	32825	Doublesex family	<2	47	129
Tcfcp213	244612	Grainyhead like 2	2.8	126	81
Osr2	46336	Odd-skipped related 2	2.1	63	0
Tcf12	171615	Transcription factor 12	<2	206	225

 TPM (tags per million) in bold are three times greater than in other normal tissue SAGE libraries (

*  except in retina;

†  except in skin). Each experiment was performed in triplicate and the data illustrated are the average of at least three experiments. Activation (x-fold) is the ratio of luciferase activity with transcription factor versus activity with the pKW10 vector, with 60 ng transcription factor DNA.

‡  These two transcription factors were combined in transfection.

Figure 2.

 

Aldh3a1 promoter constructs. The various Aldh3a1 promoter constructs used in this study are shown. The −1050/+3486 Aldh3a1 Luc construct contained 1050 bp upstream of the transcription start site (indicated by +1 and the arrow), exon 1, intron 1, and part of exon 2 (containing the initiating methionine codon of Aldh3a1) fused to the firefly Luc reporter gene. The −1050/+2986 Luc construct had a 500-bp deletion at the 3′ end of the first intron. The remaining constructs lacked the first intron and were progressively deleted from the −1050 end, as indicated.

Figure 3.

 

Synergistic activation of the Aldh3a1 promoter by Pax6, Oct1, and p300. Transient cotransfections were performed in COS-7 cells by using the −1050/+3486 Aldh3a1 Luc promoter construct with increasing amounts (15, 30, or 60 ng each) of Pax6 (A), Oct1 (B), and p300 (C) expression plasmids alone or in various combinations (C). Luciferase activity was then calculated.

Figure 4.

 

Identification of Pax6-responsive regions within the Aldh3a1 promoter. (A) Luciferase activity of various Aldh3a1 Luc promoter constructs lacking the first intron and sequential deletions at the 5′ end of the promoter was determined after cotransfection with Pax6 (30 ng) in COS-7 cells. (B) A putative Pax6 binding site at position −78 to −65 in the Aldh3a1 promoter was compared with the consensus Pax6 paired-domain binding site. Asterisks indicate identity.

Figure 5.

 

Direct binding of the Aldh3a1 promoter by Pax6 in vitro and in vivo. (A) EMSAs were performed with radiolabeled ds oligos #1 through #5, corresponding to positions −115 to −41 of the Aldh3a1 promoter or oligos #6 and #7 representing the consensus and mutated Pax6 binding site, respectively, incubated in αTN4 cell nuclear extracts. Arrow: complexes that co-migrate with the shift produced by the Pax6 consensus oligo #6. (B) EMSAs were conducted with radiolabeled oligo #6, #3, or #4 after competition with 50 or 100 ng nonradiolabeled oligo (cold competitor) as indicated or an unrelated (UR) oligo. (C) EMSA revealed immunointerference of the complex using radiolabeled oligo #6, #3, or #4 with an α-Pax6 antibody (lanes 3, 6, and 9), compared with reactions without added antibody (lanes 1, 4, and 7) or reactions with a nonspecific α-Neu Ab (lanes 2, 5, and 8). (D) PCR analysis on chromatin immunoselected with goat or rabbit α-Pax6 Ab (lanes 2 and 3, respectively) or α-Neu Ab (lane 4) using primers corresponding to positions −146 and +4 of the Aldh3a1 promoter. Lane 1: input DNA (no immunoprecipitation); lane 5: no-template-added control. MW, molecular weight ladder.

Figure 6.

 

Identification of Oct1-responsive regions within the Aldh3a1 promoter. (A) Activation of the Aldh3a1 promoter and various deletion constructs driving the luciferase reporter gene cotransfected with the Oct1 expression plasmid (30 ng) in COS-7 cells. (B) Alignment of the consensus Oct1 binding site with potential Oct1 binding sites located in the Aldh3a1 promoter. Mismatched nucleotides are underscored.

Figure 7.

 

Direct binding of the Aldh3a1 promoter by Oct1 in vitro and in vivo. (A) EMSAs were performed with radiolabeled ds oligos #9– #12 and #14, corresponding to positions +3010 to 3034, +3054 to +3078, +3311 to +3335, +3434 to +3458, and −803 to −779, respectively, of the Aldh3a1 promoter, or with oligo #8 representing the consensus Oct1-binding site incubated with HeLa cell nuclear extracts. Arrow: complexes that co-migrate with the shift produced by the Oct1 consensus oligo #8. (B) EMSAs were conducted with radiolabeled oligo #8, #12, or #14 after competition with 50 or 100 ng nonradiolabeled oligo (cold competitor) or an unrelated (UR) oligo. (C) Radiolabeled oligos #8, #12, and #14 or unrelated ds oligo #3 were incubated with α-Oct1 Ab (lanes 2, 5, 8, and 11), α-Neu Ab (lanes 3, 6, 9, and 12), or no Ab (lanes 1, 4, 7, and 10). Arrow: complexes that co-migrate with the shift produced by the Oct1 consensus oligo #8. Lower migrating bands are nonspecific, as they appeared in all lanes. (D) PCR amplification from chromatin immunoselected with α-Neu Ab (lane 2) or α-Oct1 Ab (lane 3) using primers complementary to positions −850 and −650 of the Aldh3a1 promoter. Lane 1: input DNA. (E) PCR amplification from reverse cross-linked Neu- or Oct1-immunoselected DNA (lane 2 and 3, respectively) using primers complementary to positions +3342 and +3486. Lane 1: input DNA. MW, molecular weight ladder.

Figure 8.

 

Pax6 and Oct1 localized to corneal epithelial cell nuclei. Corneal sections from 6-week-old C57Bl/6 mice were treated with no primary antibody (A, C), an antibody against Pax6 (B), or an antibody against Oct1 preincubated with a nonspecific (D, F–H) or Oct1-specific peptide (E). Arrow: Oct1 staining in keratocytes. Oct1 was localized using a secondary antibody conjugated to red Alexa Fluor 568 (F), nuclei are visualized by DAPI (blue; G). (H) Superimposition of (F) and (G) shows Oct1 localized to nuclei (arrows: purple). CE, corneal epithelium.

Figure 9.

 

Reduction in Aldh3a1 correlated with reduced Pax6 expression. In situ hybridization to examine Aldh3a1 expression was performed on corneal sections from (A) wild-type Pax6 ^SeyDey (+/+), (B) Pax6 ^SeyDey (+/−), and (C) Pax6 ^SeyNeu (+/−) mice. Q-PCR was used to quantitate the reduction in Aldh3a1 mRNA in corneas from Pax6 ^SeyDey (+/−) compared with their wild type Pax6 ^SeyDey (+/+) siblings (D). Mice ranged from 2 months to 1 year of age in each group.

Supplementary Materials

Supplementary Table S1 - (PDF) 

Supplementary Table S2 - (PDF) 

The authors thank Chun Gao for assistance with the microscope equipped with the CCD camera for epifluorescence.